T cells form intriguing patterns during adhesion to antigen-presenting cells. The patterns are composed of two types of domains, which either contain short TCR/MHCp receptor-ligand complexes or the longer LFA-1/ICAM-1 complexes. The final pattern consists of a central TCR/MHCp domain surrounded by a ring-shaped LFA-1/ICAM-1 domain, whereas the characteristic pattern formed at intermediate times is inverted with TCR/MHCp complexes at the periphery of the contact zone and LFA-1/ICAM-1 complexes in the center. Several mechanisms have been proposed to explain the T-cell pattern formation. Whereas biologists have emphasized the role of active cytoskeletal processes, previous theoretical studies suggest that the pattern evolution may be caused by spontaneous self-assembly processes alone. Some of these studies focus on circularly symmetric patterns and propose a pivot mechanism for the formation of the intermediate inverted pattern. Here, we present a statistical-mechanical model which includes thermal fluctuations and the full range of spatial patterns. We confirm the observation that the intermediate inverted pattern may be formed by spontaneous self-assembly. However, we find a different self-assembly mechanism in which numerous TCR/MHCp microdomains initially nucleate throughout the contact zone. The diffusion of free receptors and ligands into the contact zone subsequently leads to faster growth of peripheral TCR/MHCp microdomains and to a closed ring for sufficiently large TCR/MHCp concentrations. At smaller TCR/MHCp concentrations, we observe a second regime of pattern formation with characteristic multifocal intermediates, which resemble patterns observed during adhesion of immature T cells or thymozytes. In contrast to other theoretical models, we find that the final T-cell pattern with a central TCR/MHCp domain is only obtained in the presence of active cytoskeletal transport processes.